work on PackLinuxElf32arm; add nrv2b_d8.S

This commit is contained in:
John Reiser
2006-07-21 13:11:02 -07:00
parent 436532bed7
commit 52e8326d54
8 changed files with 1393 additions and 1054 deletions
File diff suppressed because it is too large Load Diff
+6 -5
View File
@@ -46,7 +46,6 @@
ldrb ip,meth; cmp ip,#M_LZMA; bne not_lzma
PUSH {fp,lr}
mov fp,sp
#define a0 r0
#define a1 r1
@@ -71,8 +70,8 @@
mov fp,fp,LSR #3 // lit_context_bits + lit_pos_bits
mov ip,ip,LSL fp // 2*LZMA_LIT_SIZE << (lit_context_bits + lit_pos_bits)
mov fp,sp
add ip,ip,# ((4*4 + 2*LZMA_BASE_SIZE)>>8)<<8
add ip,ip,#0xff & (4*4 + 2*LZMA_BASE_SIZE)
add ip,ip,# ((State + 2*LZMA_BASE_SIZE)>>8)<<8
add ip,ip,#0xff & (State + 2*LZMA_BASE_SIZE)
sub sp,sp,ip
ldr ip,[ldst]
@@ -101,6 +100,10 @@
strb ip,[sp,#0 + State]
add a0,sp,#State
bl 1f // the call
mov sp,fp
POP {fp,pc}
1:
section LZMA_DEC10
#include "lzma_d_cs.S"
@@ -109,8 +112,6 @@
#include "lzma_d_cf.S"
section LZMA_DEC30
mov sp,fp
POP {fp,pc}
not_lzma:
+121
View File
@@ -0,0 +1,121 @@
/* nrv2b_d8.S -- ARM decompressor for NRV2B
This file is part of the UPX executable compressor.
Copyright (C) 1996-2006 Markus Franz Xaver Johannes Oberhumer
Copyright (C) 1996-2006 Laszlo Molnar
Copyright (C) 2000-2006 John F. Reiser
All Rights Reserved.
UPX and the UCL library are free software; you can redistribute them
and/or modify them under the terms of the GNU General Public License as
published by the Free Software Foundation; either version 2 of
the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; see the file COPYING.
If not, write to the Free Software Foundation, Inc.,
59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
Markus F.X.J. Oberhumer Laszlo Molnar
<mfx@users.sourceforge.net> <ml1050@users.sourceforge.net>
John F. Reiser
<jreiser@users.sourceforge.net>
*/
#define src r0
#define len r1
#define dst r2
#define tmp r3
#define bits r4
#define off r5
/* macros reduce "noise" when comparing this ARM code to corresponding THUMB code */
#define PUSH stmdb sp!,
#define POP ldmia sp!,
#define ADD2( dst,src) add dst,dst,src
#define ADD2S(dst,src) adds dst,dst,src
#define ADC2( dst,src) adc dst,dst,src
#define ADC2S(dst,src) adcs dst,dst,src
#define SUB2( dst,src) sub dst,dst,src
#define SUB2S(dst,src) subs dst,dst,src
#define LDRB3(reg,psrc,incr) ldrb reg,psrc,incr
#define STRB3(reg,pdst,incr) strb reg,pdst,incr
#undef GETBIT
#define GETBIT ADD2S(bits,bits); bleq get8_n2b
#define getnextb(reg) GETBIT; ADC2S(reg,reg) /* Set Condition Codes on result */
#define jnextb0 GETBIT; bcc
#define jnextb1 GETBIT; bcs
ucl_nrv2b_decompress_8: .globl ucl_nrv2b_decompress_8 // ARM mode
.type ucl_nrv2b_decompress_8, %function
/* error = (*)(char const *src, int len_src, char *dst, int *plen_dst) */
add r1,len,src // r1= eof_src;
PUSH {r1,r2,r3, r4,r5, lr}
mvn off,#~-1 // off= -1 initial condition
mov bits,#1<<31 // refill next time
b top_n2b
eof_n2b:
POP {r1,r3,r4} // r1= eof_src; r3= orig_dst; r4= plen_dst
SUB2(src,r1) // 0 if actual src length equals expected length
SUB2(dst,r3) // actual dst length
str dst,[r4]
POP {r4,r5, pc} // return
get8_n2b: // In: Carry set [from adding 0x80000000 (1<<31) to itself]
LDRB3(bits,[src],#1) // zero-extend next byte
adc bits,bits,bits // double and insert CarryIn as low bit
movs bits,bits,lsl #24 // move to top byte, and set CarryOut from old bit 8
mov pc,lr
lit_n2b:
LDRB3(tmp,[src],#1)
STRB3(tmp,[dst],#1)
top_n2b:
jnextb1 lit_n2b
mov len,#1 // the msb
getoff_n2b: // ss11 len= [2..)
getnextb(len)
jnextb0 getoff_n2b
sub tmp,len,#3 // set Carry
mov len,#0 // Carry unaffected
blo offprev_n2b // ss11 returned 2
LDRB3(off,[src],#1) // low 8 bits
orr off,off,tmp,lsl #8
mvns off,off; beq eof_n2b // off= ~off
offprev_n2b: // In: 0==len
getnextb(len); getnextb(len); bne plus1_n2b // two bits; 1,2,3 ==> 2,3,4
mov len,#1 // the msb
getlen_n2b: // ss11 len= [2..)
getnextb(len)
jnextb0 getlen_n2b
ADD2(len,#2) // [2..) ==> [4..);
plus1_n2b:
ADD2(len,#1) // 1,2,3 ==> 2,3,4; [4..) ==> [5..)
/* 'cmn': add the inputs, set condition codes, discard the sum */
cmn off,#0xd<<8 // within M2_MAX_OFFSET
addcc len,len,#1 // too far away, so minimum match length is 3
ldrb tmp,[dst] // force cacheline allocate
copy_n2b:
ldrb tmp,[dst,off]
STRB3(tmp,[dst],#1)
SUB2S(len,#1); bne copy_n2b
b top_n2b
.size ucl_nrv2b_decompress_8, .-ucl_nrv2b_decompress_8
/*
vi:ts=8:et:nowrap
*/
+30 -22
View File
@@ -30,6 +30,7 @@
*/
#define section .section
#define bkpt .long 0xe1200070
sz_Elf32_Ehdr = 13*4
sz_Elf32_Phdr = 8*4
@@ -49,16 +50,26 @@ MAP_ANONYMOUS= 0x20
PAGE_SHIFT= 12
PAGE_SIZE = -(~0<<PAGE_SHIFT)
//.long sz_pack2 // placed there by ::pack3()
section ELFMAINX
start_params:
.long ADRM // dst for map
.long LENF // end_decompress - _start
.long CPR0 // cpr0 - _start
_start: .globl _start
/* Get some pages: enough
to duplicate the entire compressed PT_LOAD, plus 1 page, located just after
the brk() of the _un_compressed program. The address and length are pre-
calculated by PackLinuxElf32arm::pack3(), and patched in at compress time.
the brk() of the _un_compressed program. The address is pre-calculated
calculated by PackLinuxElf32arm::addLinkerSymbols().
*/
adr r12,start_params
ldmia r12!,{r0,r1, r8,r10} // ADRM,LENM, ADRC,CNTC
stmdb sp!,{r0,r1,r2} // ADRU,LENU,space for sz_unc
adr r12,start_params -4
ldmia r12!,{r1,r2, r10,r11} // r1= sz_pack2; r2= ADRM; r10= LENF; r11= CPR0;
add r10,r10,r12 // end_decompress
add r11,r11,r12 // cpr0
mov r0,r2
sub r8,r12,r1 // 4*4+ (char *)&our_Elf32_Ehdr
add r1,r1,# PAGE_SIZE
stmdb sp!,{r0,r1,r2} // ADRU,LENU,space for sz_unc
mov r2,#PROT_READ | PROT_WRITE | PROT_EXEC
mov r3,#MAP_PRIVATE | MAP_FIXED | MAP_ANONYMOUS
mvn r4,#0 // -1; cater to *BSD for fd when MAP_ANON
@@ -66,14 +77,15 @@ _start: .globl _start
cmn r0,#4096
bcs msg_SELinux
mov r9,r0 // destination for copy
bic r8,r8,#0xff // round down to page boundary
copy:
ldmia r8!,{r0,r1,r2,r3,r4,r5,r6,r7}; subs r10,r10,#1
stmia r9!,{r0,r1,r2,r3,r4,r5,r6,r7}; bne copy
ldmia r8!,{r0,r1,r2,r3,r4,r5,r6,r7}; cmp r8,r10
stmia r9!,{r0,r1,r2,r3,r4,r5,r6,r7}; blo copy
mov lr,r9 // dst for unfolded code
sub r11,r9,r8 // relocation amount
ldmia r12!,{r1,r4,r9,r10} // len_f,cpr0,LENX,ADRX
add r5,r12,r11 // relocated f_decompress
add r12,r12,# f_decompress - _start
add r5,r11,r12 // relocated f_decompress
/* linux/include/asm-arm/unistd.h */
#define __NR_SYSCALL_BASE 0x900000
@@ -81,7 +93,7 @@ copy:
#define __ARM_NR_cacheflush (__ARM_NR_BASE+2)
mov r0,r5
add r1,r1,r0 // relocated end_decompress
mov r1,r10 // relocated end_decompress
mov r2,#0
swi __ARM_NR_cacheflush // relocated decompressor
@@ -92,27 +104,23 @@ copy:
ldrb r4,[r4,# b_method ]
mov pc,r5 // decompress folded code; go there (lr)
start_params:
.long ADRM // dst for map
.long LENM // len for map
.long ADRC // src for copy
.long CNTC // cnt for copy: number of 32-byte blocks
.long len_f // length of decompressor code
.long cpr0
.long LENX // total size of compressed data
.long ADRX // &b_info of 1st compressed block (after moving)
f_decompress:
#define LINUX_ARM_CACHEFLUSH 1
section NRV_HEAD
// empty
section NRV_TAIL
// empty
section NRV2E
#include "arch/arm/v4a/nrv2e_d8.S"
section NRV2D
#include "arch/arm/v4a/nrv2d_d8.S"
section NRV2B
#include "arch/arm/v4a/nrv2b_d8.S"
#include "arch/arm/v4a/lzma_d.S"
section ELFMAINY